NaiNUQ is a neural-network-based emulator trained to reproduce the evolution of sea ice simulated by NANUQ, a numerical sea-ice model for the Arctic basin. It substantially reduces the computational cost of sea-ice simulations while maintaining high fidelity to the original model.

Designed for ensemble experiments, uncertainty quantification, and data-assimilation workflows, NaiNUQ is available at four temporal resolutions (1 h, 6 h, 12 h, and 24 h) and includes pre-trained model weights ready for use.